lets you control penis depth with the lora value,
0~1 recomended for vaginal and anal
0~X recomended for deepthroat
deepthroat is more sensitive to the lora, anywhere from 0.2 to 0.4 might be the same as all the way to the base, but you can put it higher, if you do, then it trends to top-down bottom-up
mainly trained on pov but it also works okay for from side
Description
this is the five epoch version, minimal effect on style, but doesn't combine well with style loras
FAQ
Comments (36)
Edit: or not, I think I might have imagined the style leak? Let me know what you get. v1.0, the 12 epoch version seems to be superior in every way.
NOTE! There's two versions up, v1.0 is the 12 epoch version, v1.0-000005 is the 5 epoch version.
The five epoch version works on its own fine, but not well with other loras.
The twelve epoch version works well with other loras, but on its own, it has a noticeable, not huge, but noticeable impact on style.
Does this not go in the LORA folder or does it only work when using Pony?
yeah, it goes in your lora folder, you use it like you use any lora.
it'll function with other models, but I didn't test it on any other than pony XL, so I'm not sure how good it will do.
hi, this Lora is not really seen by my stable diffusion. How come?
liking my results with this on. im curious what goes into a concept dataset so you dont have stylebleed? i was thinking of using a bunch of different artists with the desired concept or an outlandish idea of maybe slightly blurring the images
if you blur the images, you'll want to make sure you tag them that they're blurred, and lower the score_# for the blureed versions in your data set. That might work.
For this particular lora; I trained it on 270 images of vaginal anal and deepthroat I think that helps in avoiding style bleed, that it's a single concept linked through three different poses, and a whole bunch of different artists.
also lots of training, this one I think was around 10k steps
i cant seem to get the same effect with dildos T_T
My favorite Lora for dephttoat
and remember: the deeper she goes the more she loves you ! :D lmao
Any chance for a SD 1.5 version?
Sorry for my ignorance, but how do I achieve the gif format?
generate one image at a time, then use python, or a website, or some other third party tool to convert them into a gif. I use python. You can easily get what you need from chatgpt; just ask it to write a python script to convert a folder full of images into a gif.
@emberbyte well that might be too much for me, anyway how did you make the image like 4 in one?nd you used seed to consistency right? And what's the height for the Lora you recommend
@kd231002573 that's stuff I did with python; I wrote scripts to create gifs, and to combine them.
For consistency, yeah, I used the same seed, and this lora with varried strenght.
@emberbyte Hey, How did you manage to get matching frames? Did you use one seed or did you use animatediff? Animatediff seems not to work for me with the pony model
Any idea why my generations come without a a pen1s at all or this lora just doesn't work?
Some checkpoint just do not like the idea of deepthroating at all. It is not LORA fault.
Excellent LoRA. More people should see this.
12 epoch version is listed as non-Pony but the examples have the Pony prompt. Can you confirm the 12 epoch version is NOT base SDXL and IS Pony-trained?
this unreal huge cock kills the vibe. is there a way to control the cock size?
Try to avoid using words like big, thick etc. in your prompt
Love it! This is the best one for this effect ;)
dont get clip 1 with pony which uses clip skip -2, and can´t get deep anal to work in the slightest like in the preview pics
best masterwork lora ever.
amazing and greatly needed
specialized to the pov, but VERY GOOD at adding depth to penetration.
Fuck dude you're my savior, i've been looking for this for a while, excellent work, keep it up!
worked at IllustriousXL. nice.
Never seems to go down to the base for me. I tried deepthroat 0.2 all the way up 40
god bless you